MULTISPECTRAL AND SAR IMAGE FUSION BASED ON LAPLACIAN PYRAMID AND SPARSE REPRESENTATION

摘要

Multispectral images contain plentiful spectral information of the Earth's surface, which is beneficial for identifying land cover types. While Synthetic Aperture Radar (SAR) images can provide abundant information on texture and structure of target objects. Complementary information from multi-sensors can be combined to improve the availability and reliability of stand-alone dataset. Therefore, this paper presents a fusion framework based on Laplacian pyramid (LP) and sparse representation theory to integrate information from multispectral and SAR images. Laplacian pyramid is performed to decompose both the multispectral and SAR images into multi-level detail images (or high-frequency components) and approximation images (or low-frequency components). High-frequency components are combined based on the absolute values of coefficients while low-frequency components are merged based on sparse representation theory. Finally. LP reconstruction is performed on fused high-frequency and low-frequency components to obtain the integrated image. We conduct experiments on several datasets to verify the effectiveness of proposed method. Both visual interpretation and statistical analyses demonstrate that the proposed method strike a satisfactory balance between spectral information preservation and enhancement of spatial and textual characteristics.